Add Training Data
Browse files- trainer_state.json +1208 -0
- training_args.bin +3 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1208 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.8,
|
| 5 |
+
"eval_steps": 12,
|
| 6 |
+
"global_step": 60,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.08,
|
| 13 |
+
"grad_norm": 5.910620212554932,
|
| 14 |
+
"learning_rate": 8e-07,
|
| 15 |
+
"log_odds_chosen": -0.012692844495177269,
|
| 16 |
+
"log_odds_ratio": -0.7021465301513672,
|
| 17 |
+
"logits/chosen": -2.7243709564208984,
|
| 18 |
+
"logits/rejected": -2.7299304008483887,
|
| 19 |
+
"logps/chosen": -1.1940056085586548,
|
| 20 |
+
"logps/rejected": -1.1935062408447266,
|
| 21 |
+
"loss": 1.4594,
|
| 22 |
+
"nll_loss": 1.3891654014587402,
|
| 23 |
+
"rewards/accuracies": 0.25,
|
| 24 |
+
"rewards/chosen": -0.11940056085586548,
|
| 25 |
+
"rewards/margins": -4.9931928515434265e-05,
|
| 26 |
+
"rewards/rejected": -0.11935063451528549,
|
| 27 |
+
"step": 1
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.16,
|
| 31 |
+
"grad_norm": 5.268943786621094,
|
| 32 |
+
"learning_rate": 1.6e-06,
|
| 33 |
+
"log_odds_chosen": 0.22696708142757416,
|
| 34 |
+
"log_odds_ratio": -0.643199622631073,
|
| 35 |
+
"logits/chosen": -2.8990988731384277,
|
| 36 |
+
"logits/rejected": -2.82047700881958,
|
| 37 |
+
"logps/chosen": -1.2680704593658447,
|
| 38 |
+
"logps/rejected": -1.4174962043762207,
|
| 39 |
+
"loss": 1.6127,
|
| 40 |
+
"nll_loss": 1.5484061241149902,
|
| 41 |
+
"rewards/accuracies": 0.75,
|
| 42 |
+
"rewards/chosen": -0.1268070489168167,
|
| 43 |
+
"rewards/margins": 0.014942573383450508,
|
| 44 |
+
"rewards/rejected": -0.14174960553646088,
|
| 45 |
+
"step": 2
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.24,
|
| 49 |
+
"grad_norm": 4.226258277893066,
|
| 50 |
+
"learning_rate": 2.4e-06,
|
| 51 |
+
"log_odds_chosen": 0.044683486223220825,
|
| 52 |
+
"log_odds_ratio": -0.7104766368865967,
|
| 53 |
+
"logits/chosen": -2.8773069381713867,
|
| 54 |
+
"logits/rejected": -2.9280641078948975,
|
| 55 |
+
"logps/chosen": -1.150111198425293,
|
| 56 |
+
"logps/rejected": -1.1700233221054077,
|
| 57 |
+
"loss": 1.6466,
|
| 58 |
+
"nll_loss": 1.575596809387207,
|
| 59 |
+
"rewards/accuracies": 0.25,
|
| 60 |
+
"rewards/chosen": -0.11501111835241318,
|
| 61 |
+
"rewards/margins": 0.001991217490285635,
|
| 62 |
+
"rewards/rejected": -0.11700233817100525,
|
| 63 |
+
"step": 3
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.32,
|
| 67 |
+
"grad_norm": 4.236293792724609,
|
| 68 |
+
"learning_rate": 3.2e-06,
|
| 69 |
+
"log_odds_chosen": 0.11046111583709717,
|
| 70 |
+
"log_odds_ratio": -0.6462637782096863,
|
| 71 |
+
"logits/chosen": -3.1065869331359863,
|
| 72 |
+
"logits/rejected": -2.97692608833313,
|
| 73 |
+
"logps/chosen": -1.0575246810913086,
|
| 74 |
+
"logps/rejected": -1.1389795541763306,
|
| 75 |
+
"loss": 1.3322,
|
| 76 |
+
"nll_loss": 1.2675998210906982,
|
| 77 |
+
"rewards/accuracies": 0.5,
|
| 78 |
+
"rewards/chosen": -0.10575246810913086,
|
| 79 |
+
"rewards/margins": 0.008145488798618317,
|
| 80 |
+
"rewards/rejected": -0.11389796435832977,
|
| 81 |
+
"step": 4
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.4,
|
| 85 |
+
"grad_norm": 4.288541316986084,
|
| 86 |
+
"learning_rate": 4e-06,
|
| 87 |
+
"log_odds_chosen": 0.2469336837530136,
|
| 88 |
+
"log_odds_ratio": -0.6131026148796082,
|
| 89 |
+
"logits/chosen": -3.0275461673736572,
|
| 90 |
+
"logits/rejected": -3.003474473953247,
|
| 91 |
+
"logps/chosen": -0.9929811954498291,
|
| 92 |
+
"logps/rejected": -1.1053597927093506,
|
| 93 |
+
"loss": 1.3442,
|
| 94 |
+
"nll_loss": 1.2829171419143677,
|
| 95 |
+
"rewards/accuracies": 0.75,
|
| 96 |
+
"rewards/chosen": -0.09929811954498291,
|
| 97 |
+
"rewards/margins": 0.011237863451242447,
|
| 98 |
+
"rewards/rejected": -0.11053598672151566,
|
| 99 |
+
"step": 5
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.48,
|
| 103 |
+
"grad_norm": 4.556058883666992,
|
| 104 |
+
"learning_rate": 4.8e-06,
|
| 105 |
+
"log_odds_chosen": 0.048641495406627655,
|
| 106 |
+
"log_odds_ratio": -0.691319465637207,
|
| 107 |
+
"logits/chosen": -2.9148268699645996,
|
| 108 |
+
"logits/rejected": -2.8536345958709717,
|
| 109 |
+
"logps/chosen": -1.3467779159545898,
|
| 110 |
+
"logps/rejected": -1.3747429847717285,
|
| 111 |
+
"loss": 1.4166,
|
| 112 |
+
"nll_loss": 1.3474537134170532,
|
| 113 |
+
"rewards/accuracies": 0.5,
|
| 114 |
+
"rewards/chosen": -0.13467779755592346,
|
| 115 |
+
"rewards/margins": 0.0027965055778622627,
|
| 116 |
+
"rewards/rejected": -0.13747431337833405,
|
| 117 |
+
"step": 6
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.56,
|
| 121 |
+
"grad_norm": 4.545846462249756,
|
| 122 |
+
"learning_rate": 5.6e-06,
|
| 123 |
+
"log_odds_chosen": 0.32562392950057983,
|
| 124 |
+
"log_odds_ratio": -0.5898037552833557,
|
| 125 |
+
"logits/chosen": -2.815234899520874,
|
| 126 |
+
"logits/rejected": -2.7621359825134277,
|
| 127 |
+
"logps/chosen": -1.061276912689209,
|
| 128 |
+
"logps/rejected": -1.2360599040985107,
|
| 129 |
+
"loss": 1.3822,
|
| 130 |
+
"nll_loss": 1.3232545852661133,
|
| 131 |
+
"rewards/accuracies": 0.625,
|
| 132 |
+
"rewards/chosen": -0.10612770169973373,
|
| 133 |
+
"rewards/margins": 0.017478302121162415,
|
| 134 |
+
"rewards/rejected": -0.12360600382089615,
|
| 135 |
+
"step": 7
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.64,
|
| 139 |
+
"grad_norm": 3.971618890762329,
|
| 140 |
+
"learning_rate": 6.4e-06,
|
| 141 |
+
"log_odds_chosen": 0.1728549599647522,
|
| 142 |
+
"log_odds_ratio": -0.6392964720726013,
|
| 143 |
+
"logits/chosen": -2.9990975856781006,
|
| 144 |
+
"logits/rejected": -2.9307050704956055,
|
| 145 |
+
"logps/chosen": -1.190705418586731,
|
| 146 |
+
"logps/rejected": -1.3564025163650513,
|
| 147 |
+
"loss": 1.5385,
|
| 148 |
+
"nll_loss": 1.4745732545852661,
|
| 149 |
+
"rewards/accuracies": 0.375,
|
| 150 |
+
"rewards/chosen": -0.11907054483890533,
|
| 151 |
+
"rewards/margins": 0.016569709405303,
|
| 152 |
+
"rewards/rejected": -0.1356402486562729,
|
| 153 |
+
"step": 8
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.72,
|
| 157 |
+
"grad_norm": 2.60530686378479,
|
| 158 |
+
"learning_rate": 7.2e-06,
|
| 159 |
+
"log_odds_chosen": 0.17511004209518433,
|
| 160 |
+
"log_odds_ratio": -0.63337641954422,
|
| 161 |
+
"logits/chosen": -3.026963233947754,
|
| 162 |
+
"logits/rejected": -2.9191253185272217,
|
| 163 |
+
"logps/chosen": -1.0979580879211426,
|
| 164 |
+
"logps/rejected": -1.2179211378097534,
|
| 165 |
+
"loss": 1.2909,
|
| 166 |
+
"nll_loss": 1.2275505065917969,
|
| 167 |
+
"rewards/accuracies": 0.625,
|
| 168 |
+
"rewards/chosen": -0.10979580879211426,
|
| 169 |
+
"rewards/margins": 0.011996296234428883,
|
| 170 |
+
"rewards/rejected": -0.12179210036993027,
|
| 171 |
+
"step": 9
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.8,
|
| 175 |
+
"grad_norm": 1.7484970092773438,
|
| 176 |
+
"learning_rate": 8e-06,
|
| 177 |
+
"log_odds_chosen": 0.2105366289615631,
|
| 178 |
+
"log_odds_ratio": -0.6329271793365479,
|
| 179 |
+
"logits/chosen": -2.628000259399414,
|
| 180 |
+
"logits/rejected": -2.4390053749084473,
|
| 181 |
+
"logps/chosen": -0.9455797076225281,
|
| 182 |
+
"logps/rejected": -1.1043422222137451,
|
| 183 |
+
"loss": 1.1688,
|
| 184 |
+
"nll_loss": 1.1055198907852173,
|
| 185 |
+
"rewards/accuracies": 0.375,
|
| 186 |
+
"rewards/chosen": -0.09455796331167221,
|
| 187 |
+
"rewards/margins": 0.015876244753599167,
|
| 188 |
+
"rewards/rejected": -0.11043421179056168,
|
| 189 |
+
"step": 10
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.88,
|
| 193 |
+
"grad_norm": 1.6517988443374634,
|
| 194 |
+
"learning_rate": 7.84e-06,
|
| 195 |
+
"log_odds_chosen": 0.2828138470649719,
|
| 196 |
+
"log_odds_ratio": -0.5968219041824341,
|
| 197 |
+
"logits/chosen": -3.0771772861480713,
|
| 198 |
+
"logits/rejected": -3.008608341217041,
|
| 199 |
+
"logps/chosen": -0.6074647903442383,
|
| 200 |
+
"logps/rejected": -0.7685144543647766,
|
| 201 |
+
"loss": 1.1134,
|
| 202 |
+
"nll_loss": 1.053712010383606,
|
| 203 |
+
"rewards/accuracies": 0.625,
|
| 204 |
+
"rewards/chosen": -0.060746483504772186,
|
| 205 |
+
"rewards/margins": 0.016104966402053833,
|
| 206 |
+
"rewards/rejected": -0.07685144990682602,
|
| 207 |
+
"step": 11
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.96,
|
| 211 |
+
"grad_norm": 2.5685229301452637,
|
| 212 |
+
"learning_rate": 7.68e-06,
|
| 213 |
+
"log_odds_chosen": 0.12323664128780365,
|
| 214 |
+
"log_odds_ratio": -0.6441441774368286,
|
| 215 |
+
"logits/chosen": -2.9566874504089355,
|
| 216 |
+
"logits/rejected": -2.8413052558898926,
|
| 217 |
+
"logps/chosen": -1.161773443222046,
|
| 218 |
+
"logps/rejected": -1.2560498714447021,
|
| 219 |
+
"loss": 1.3902,
|
| 220 |
+
"nll_loss": 1.3257489204406738,
|
| 221 |
+
"rewards/accuracies": 0.625,
|
| 222 |
+
"rewards/chosen": -0.11617735028266907,
|
| 223 |
+
"rewards/margins": 0.009427637793123722,
|
| 224 |
+
"rewards/rejected": -0.12560498714447021,
|
| 225 |
+
"step": 12
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.96,
|
| 229 |
+
"eval_log_odds_chosen": 0.18088191747665405,
|
| 230 |
+
"eval_log_odds_ratio": -0.6067904233932495,
|
| 231 |
+
"eval_logits/chosen": -3.32541561126709,
|
| 232 |
+
"eval_logits/rejected": -2.9764161109924316,
|
| 233 |
+
"eval_logps/chosen": -0.7066247463226318,
|
| 234 |
+
"eval_logps/rejected": -0.8023566603660583,
|
| 235 |
+
"eval_loss": 0.9977810382843018,
|
| 236 |
+
"eval_nll_loss": 0.9371020197868347,
|
| 237 |
+
"eval_rewards/accuracies": 1.0,
|
| 238 |
+
"eval_rewards/chosen": -0.0706624761223793,
|
| 239 |
+
"eval_rewards/margins": 0.009573191404342651,
|
| 240 |
+
"eval_rewards/rejected": -0.08023566752672195,
|
| 241 |
+
"eval_runtime": 4.0974,
|
| 242 |
+
"eval_samples_per_second": 0.244,
|
| 243 |
+
"eval_steps_per_second": 0.244,
|
| 244 |
+
"step": 12
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 1.04,
|
| 248 |
+
"grad_norm": 2.8649556636810303,
|
| 249 |
+
"learning_rate": 7.519999999999999e-06,
|
| 250 |
+
"log_odds_chosen": 0.7436250448226929,
|
| 251 |
+
"log_odds_ratio": -0.42550715804100037,
|
| 252 |
+
"logits/chosen": -3.073974847793579,
|
| 253 |
+
"logits/rejected": -2.865658760070801,
|
| 254 |
+
"logps/chosen": -1.1165134906768799,
|
| 255 |
+
"logps/rejected": -1.6563222408294678,
|
| 256 |
+
"loss": 1.2329,
|
| 257 |
+
"nll_loss": 1.190321922302246,
|
| 258 |
+
"rewards/accuracies": 0.875,
|
| 259 |
+
"rewards/chosen": -0.11165135353803635,
|
| 260 |
+
"rewards/margins": 0.05398087203502655,
|
| 261 |
+
"rewards/rejected": -0.1656322181224823,
|
| 262 |
+
"step": 13
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 1.12,
|
| 266 |
+
"grad_norm": 1.9388059377670288,
|
| 267 |
+
"learning_rate": 7.36e-06,
|
| 268 |
+
"log_odds_chosen": 0.4244161546230316,
|
| 269 |
+
"log_odds_ratio": -0.5222975015640259,
|
| 270 |
+
"logits/chosen": -2.839306354522705,
|
| 271 |
+
"logits/rejected": -2.7786505222320557,
|
| 272 |
+
"logps/chosen": -0.7469608783721924,
|
| 273 |
+
"logps/rejected": -0.990905225276947,
|
| 274 |
+
"loss": 1.0928,
|
| 275 |
+
"nll_loss": 1.0406169891357422,
|
| 276 |
+
"rewards/accuracies": 0.75,
|
| 277 |
+
"rewards/chosen": -0.07469607889652252,
|
| 278 |
+
"rewards/margins": 0.02439444325864315,
|
| 279 |
+
"rewards/rejected": -0.09909053146839142,
|
| 280 |
+
"step": 14
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.2,
|
| 284 |
+
"grad_norm": 2.8289878368377686,
|
| 285 |
+
"learning_rate": 7.2e-06,
|
| 286 |
+
"log_odds_chosen": 0.02547745406627655,
|
| 287 |
+
"log_odds_ratio": -0.7126201391220093,
|
| 288 |
+
"logits/chosen": -2.8896398544311523,
|
| 289 |
+
"logits/rejected": -2.7489066123962402,
|
| 290 |
+
"logps/chosen": -1.080715537071228,
|
| 291 |
+
"logps/rejected": -1.107945442199707,
|
| 292 |
+
"loss": 1.4342,
|
| 293 |
+
"nll_loss": 1.3629157543182373,
|
| 294 |
+
"rewards/accuracies": 0.5,
|
| 295 |
+
"rewards/chosen": -0.10807155072689056,
|
| 296 |
+
"rewards/margins": 0.0027229972183704376,
|
| 297 |
+
"rewards/rejected": -0.1107945516705513,
|
| 298 |
+
"step": 15
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 1.28,
|
| 302 |
+
"grad_norm": 2.7624049186706543,
|
| 303 |
+
"learning_rate": 7.0399999999999995e-06,
|
| 304 |
+
"log_odds_chosen": 0.18877235054969788,
|
| 305 |
+
"log_odds_ratio": -0.6148291826248169,
|
| 306 |
+
"logits/chosen": -2.8152570724487305,
|
| 307 |
+
"logits/rejected": -2.7555055618286133,
|
| 308 |
+
"logps/chosen": -1.103430986404419,
|
| 309 |
+
"logps/rejected": -1.2451709508895874,
|
| 310 |
+
"loss": 1.3929,
|
| 311 |
+
"nll_loss": 1.3314636945724487,
|
| 312 |
+
"rewards/accuracies": 0.5,
|
| 313 |
+
"rewards/chosen": -0.1103430837392807,
|
| 314 |
+
"rewards/margins": 0.014174011535942554,
|
| 315 |
+
"rewards/rejected": -0.12451709806919098,
|
| 316 |
+
"step": 16
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 1.3599999999999999,
|
| 320 |
+
"grad_norm": 2.5873303413391113,
|
| 321 |
+
"learning_rate": 6.879999999999999e-06,
|
| 322 |
+
"log_odds_chosen": -0.10925386846065521,
|
| 323 |
+
"log_odds_ratio": -0.7573699355125427,
|
| 324 |
+
"logits/chosen": -3.056697368621826,
|
| 325 |
+
"logits/rejected": -3.0701730251312256,
|
| 326 |
+
"logps/chosen": -1.0829575061798096,
|
| 327 |
+
"logps/rejected": -1.009861946105957,
|
| 328 |
+
"loss": 1.3078,
|
| 329 |
+
"nll_loss": 1.2320821285247803,
|
| 330 |
+
"rewards/accuracies": 0.375,
|
| 331 |
+
"rewards/chosen": -0.1082957535982132,
|
| 332 |
+
"rewards/margins": -0.007309555076062679,
|
| 333 |
+
"rewards/rejected": -0.10098619759082794,
|
| 334 |
+
"step": 17
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 1.44,
|
| 338 |
+
"grad_norm": 1.7549221515655518,
|
| 339 |
+
"learning_rate": 6.719999999999999e-06,
|
| 340 |
+
"log_odds_chosen": 0.25229498744010925,
|
| 341 |
+
"log_odds_ratio": -0.605067253112793,
|
| 342 |
+
"logits/chosen": -3.0717437267303467,
|
| 343 |
+
"logits/rejected": -3.031426191329956,
|
| 344 |
+
"logps/chosen": -0.6300977468490601,
|
| 345 |
+
"logps/rejected": -0.7678447961807251,
|
| 346 |
+
"loss": 1.1493,
|
| 347 |
+
"nll_loss": 1.0887819528579712,
|
| 348 |
+
"rewards/accuracies": 0.5,
|
| 349 |
+
"rewards/chosen": -0.06300977617502213,
|
| 350 |
+
"rewards/margins": 0.013774698600172997,
|
| 351 |
+
"rewards/rejected": -0.07678447663784027,
|
| 352 |
+
"step": 18
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 1.52,
|
| 356 |
+
"grad_norm": 2.258375644683838,
|
| 357 |
+
"learning_rate": 6.559999999999999e-06,
|
| 358 |
+
"log_odds_chosen": 0.09016017615795135,
|
| 359 |
+
"log_odds_ratio": -0.6553211212158203,
|
| 360 |
+
"logits/chosen": -2.851407766342163,
|
| 361 |
+
"logits/rejected": -2.8655269145965576,
|
| 362 |
+
"logps/chosen": -1.025189995765686,
|
| 363 |
+
"logps/rejected": -1.0774447917938232,
|
| 364 |
+
"loss": 1.2653,
|
| 365 |
+
"nll_loss": 1.1997864246368408,
|
| 366 |
+
"rewards/accuracies": 0.5,
|
| 367 |
+
"rewards/chosen": -0.10251899808645248,
|
| 368 |
+
"rewards/margins": 0.005225480534136295,
|
| 369 |
+
"rewards/rejected": -0.10774447023868561,
|
| 370 |
+
"step": 19
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 1.6,
|
| 374 |
+
"grad_norm": 1.6248959302902222,
|
| 375 |
+
"learning_rate": 6.4e-06,
|
| 376 |
+
"log_odds_chosen": 0.22705766558647156,
|
| 377 |
+
"log_odds_ratio": -0.6247647404670715,
|
| 378 |
+
"logits/chosen": -2.6382391452789307,
|
| 379 |
+
"logits/rejected": -2.520052671432495,
|
| 380 |
+
"logps/chosen": -0.8028525710105896,
|
| 381 |
+
"logps/rejected": -0.9037848114967346,
|
| 382 |
+
"loss": 0.9198,
|
| 383 |
+
"nll_loss": 0.8573005199432373,
|
| 384 |
+
"rewards/accuracies": 0.375,
|
| 385 |
+
"rewards/chosen": -0.08028525859117508,
|
| 386 |
+
"rewards/margins": 0.010093222372233868,
|
| 387 |
+
"rewards/rejected": -0.09037847816944122,
|
| 388 |
+
"step": 20
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.6800000000000002,
|
| 392 |
+
"grad_norm": 2.189142942428589,
|
| 393 |
+
"learning_rate": 6.2399999999999995e-06,
|
| 394 |
+
"log_odds_chosen": 0.1090141013264656,
|
| 395 |
+
"log_odds_ratio": -0.6825957894325256,
|
| 396 |
+
"logits/chosen": -2.9855175018310547,
|
| 397 |
+
"logits/rejected": -2.871065855026245,
|
| 398 |
+
"logps/chosen": -0.9656553268432617,
|
| 399 |
+
"logps/rejected": -1.0403013229370117,
|
| 400 |
+
"loss": 1.2366,
|
| 401 |
+
"nll_loss": 1.1683895587921143,
|
| 402 |
+
"rewards/accuracies": 0.5,
|
| 403 |
+
"rewards/chosen": -0.09656553715467453,
|
| 404 |
+
"rewards/margins": 0.007464595139026642,
|
| 405 |
+
"rewards/rejected": -0.10403012484312057,
|
| 406 |
+
"step": 21
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 1.76,
|
| 410 |
+
"grad_norm": 2.530574083328247,
|
| 411 |
+
"learning_rate": 6.079999999999999e-06,
|
| 412 |
+
"log_odds_chosen": 0.11634103953838348,
|
| 413 |
+
"log_odds_ratio": -0.6487681269645691,
|
| 414 |
+
"logits/chosen": -3.062095880508423,
|
| 415 |
+
"logits/rejected": -2.9822239875793457,
|
| 416 |
+
"logps/chosen": -0.9827238321304321,
|
| 417 |
+
"logps/rejected": -1.0462595224380493,
|
| 418 |
+
"loss": 1.0995,
|
| 419 |
+
"nll_loss": 1.0346102714538574,
|
| 420 |
+
"rewards/accuracies": 0.625,
|
| 421 |
+
"rewards/chosen": -0.09827239066362381,
|
| 422 |
+
"rewards/margins": 0.006353557575494051,
|
| 423 |
+
"rewards/rejected": -0.10462595522403717,
|
| 424 |
+
"step": 22
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 1.8399999999999999,
|
| 428 |
+
"grad_norm": 1.5682123899459839,
|
| 429 |
+
"learning_rate": 5.92e-06,
|
| 430 |
+
"log_odds_chosen": 0.05784512683749199,
|
| 431 |
+
"log_odds_ratio": -0.686516284942627,
|
| 432 |
+
"logits/chosen": -3.078280210494995,
|
| 433 |
+
"logits/rejected": -3.0346872806549072,
|
| 434 |
+
"logps/chosen": -0.8539592623710632,
|
| 435 |
+
"logps/rejected": -0.9186102151870728,
|
| 436 |
+
"loss": 1.2162,
|
| 437 |
+
"nll_loss": 1.1475117206573486,
|
| 438 |
+
"rewards/accuracies": 0.375,
|
| 439 |
+
"rewards/chosen": -0.0853959321975708,
|
| 440 |
+
"rewards/margins": 0.006465096492320299,
|
| 441 |
+
"rewards/rejected": -0.09186102449893951,
|
| 442 |
+
"step": 23
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.92,
|
| 446 |
+
"grad_norm": 2.3258304595947266,
|
| 447 |
+
"learning_rate": 5.76e-06,
|
| 448 |
+
"log_odds_chosen": 0.18300886452198029,
|
| 449 |
+
"log_odds_ratio": -0.6389400959014893,
|
| 450 |
+
"logits/chosen": -3.1072592735290527,
|
| 451 |
+
"logits/rejected": -2.9962942600250244,
|
| 452 |
+
"logps/chosen": -1.1144001483917236,
|
| 453 |
+
"logps/rejected": -1.2568085193634033,
|
| 454 |
+
"loss": 1.5199,
|
| 455 |
+
"nll_loss": 1.4559848308563232,
|
| 456 |
+
"rewards/accuracies": 0.75,
|
| 457 |
+
"rewards/chosen": -0.111440010368824,
|
| 458 |
+
"rewards/margins": 0.014240838587284088,
|
| 459 |
+
"rewards/rejected": -0.1256808340549469,
|
| 460 |
+
"step": 24
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 1.92,
|
| 464 |
+
"eval_log_odds_chosen": 0.09352552890777588,
|
| 465 |
+
"eval_log_odds_ratio": -0.6474775075912476,
|
| 466 |
+
"eval_logits/chosen": -3.3521740436553955,
|
| 467 |
+
"eval_logits/rejected": -3.045846700668335,
|
| 468 |
+
"eval_logps/chosen": -0.6791949272155762,
|
| 469 |
+
"eval_logps/rejected": -0.7263939380645752,
|
| 470 |
+
"eval_loss": 0.958454430103302,
|
| 471 |
+
"eval_nll_loss": 0.8937066793441772,
|
| 472 |
+
"eval_rewards/accuracies": 1.0,
|
| 473 |
+
"eval_rewards/chosen": -0.06791949272155762,
|
| 474 |
+
"eval_rewards/margins": 0.0047199055552482605,
|
| 475 |
+
"eval_rewards/rejected": -0.07263939827680588,
|
| 476 |
+
"eval_runtime": 3.9151,
|
| 477 |
+
"eval_samples_per_second": 0.255,
|
| 478 |
+
"eval_steps_per_second": 0.255,
|
| 479 |
+
"step": 24
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 2.0,
|
| 483 |
+
"grad_norm": 1.6808022260665894,
|
| 484 |
+
"learning_rate": 5.6e-06,
|
| 485 |
+
"log_odds_chosen": -0.04988832771778107,
|
| 486 |
+
"log_odds_ratio": -0.7225002646446228,
|
| 487 |
+
"logits/chosen": -3.167830467224121,
|
| 488 |
+
"logits/rejected": -3.061304807662964,
|
| 489 |
+
"logps/chosen": -0.8571638464927673,
|
| 490 |
+
"logps/rejected": -0.827689528465271,
|
| 491 |
+
"loss": 1.2537,
|
| 492 |
+
"nll_loss": 1.1814517974853516,
|
| 493 |
+
"rewards/accuracies": 0.375,
|
| 494 |
+
"rewards/chosen": -0.08571638911962509,
|
| 495 |
+
"rewards/margins": -0.002947437111288309,
|
| 496 |
+
"rewards/rejected": -0.08276894688606262,
|
| 497 |
+
"step": 25
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 2.08,
|
| 501 |
+
"grad_norm": 2.2546420097351074,
|
| 502 |
+
"learning_rate": 5.4400000000000004e-06,
|
| 503 |
+
"log_odds_chosen": 0.04561331868171692,
|
| 504 |
+
"log_odds_ratio": -0.7332349419593811,
|
| 505 |
+
"logits/chosen": -2.59240460395813,
|
| 506 |
+
"logits/rejected": -2.582451343536377,
|
| 507 |
+
"logps/chosen": -0.9324518442153931,
|
| 508 |
+
"logps/rejected": -0.953067421913147,
|
| 509 |
+
"loss": 1.1664,
|
| 510 |
+
"nll_loss": 1.0930918455123901,
|
| 511 |
+
"rewards/accuracies": 0.5,
|
| 512 |
+
"rewards/chosen": -0.09324519336223602,
|
| 513 |
+
"rewards/margins": 0.0020615560933947563,
|
| 514 |
+
"rewards/rejected": -0.09530673921108246,
|
| 515 |
+
"step": 26
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 2.16,
|
| 519 |
+
"grad_norm": 2.0570733547210693,
|
| 520 |
+
"learning_rate": 5.28e-06,
|
| 521 |
+
"log_odds_chosen": 0.24187083542346954,
|
| 522 |
+
"log_odds_ratio": -0.6241340637207031,
|
| 523 |
+
"logits/chosen": -3.080639600753784,
|
| 524 |
+
"logits/rejected": -3.0304229259490967,
|
| 525 |
+
"logps/chosen": -0.7420560121536255,
|
| 526 |
+
"logps/rejected": -0.8350604176521301,
|
| 527 |
+
"loss": 1.0571,
|
| 528 |
+
"nll_loss": 0.9946718811988831,
|
| 529 |
+
"rewards/accuracies": 0.5,
|
| 530 |
+
"rewards/chosen": -0.07420559972524643,
|
| 531 |
+
"rewards/margins": 0.009300438687205315,
|
| 532 |
+
"rewards/rejected": -0.0835060402750969,
|
| 533 |
+
"step": 27
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 2.24,
|
| 537 |
+
"grad_norm": 2.023972511291504,
|
| 538 |
+
"learning_rate": 5.12e-06,
|
| 539 |
+
"log_odds_chosen": 0.19169293344020844,
|
| 540 |
+
"log_odds_ratio": -0.6201201677322388,
|
| 541 |
+
"logits/chosen": -2.8682751655578613,
|
| 542 |
+
"logits/rejected": -2.872384548187256,
|
| 543 |
+
"logps/chosen": -0.8573394417762756,
|
| 544 |
+
"logps/rejected": -1.0006366968154907,
|
| 545 |
+
"loss": 1.1392,
|
| 546 |
+
"nll_loss": 1.077183723449707,
|
| 547 |
+
"rewards/accuracies": 0.375,
|
| 548 |
+
"rewards/chosen": -0.08573393523693085,
|
| 549 |
+
"rewards/margins": 0.014329724945127964,
|
| 550 |
+
"rewards/rejected": -0.10006366670131683,
|
| 551 |
+
"step": 28
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 2.32,
|
| 555 |
+
"grad_norm": 2.2322771549224854,
|
| 556 |
+
"learning_rate": 4.96e-06,
|
| 557 |
+
"log_odds_chosen": 0.22571854293346405,
|
| 558 |
+
"log_odds_ratio": -0.6048489212989807,
|
| 559 |
+
"logits/chosen": -2.7974281311035156,
|
| 560 |
+
"logits/rejected": -2.671107292175293,
|
| 561 |
+
"logps/chosen": -0.9511339068412781,
|
| 562 |
+
"logps/rejected": -1.108659267425537,
|
| 563 |
+
"loss": 1.2377,
|
| 564 |
+
"nll_loss": 1.1771889925003052,
|
| 565 |
+
"rewards/accuracies": 0.625,
|
| 566 |
+
"rewards/chosen": -0.09511338919401169,
|
| 567 |
+
"rewards/margins": 0.015752539038658142,
|
| 568 |
+
"rewards/rejected": -0.11086593568325043,
|
| 569 |
+
"step": 29
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 2.4,
|
| 573 |
+
"grad_norm": 1.5203155279159546,
|
| 574 |
+
"learning_rate": 4.8e-06,
|
| 575 |
+
"log_odds_chosen": 0.08203547447919846,
|
| 576 |
+
"log_odds_ratio": -0.6810359358787537,
|
| 577 |
+
"logits/chosen": -3.129624366760254,
|
| 578 |
+
"logits/rejected": -3.040184497833252,
|
| 579 |
+
"logps/chosen": -1.0037201642990112,
|
| 580 |
+
"logps/rejected": -1.065619945526123,
|
| 581 |
+
"loss": 1.2065,
|
| 582 |
+
"nll_loss": 1.1384248733520508,
|
| 583 |
+
"rewards/accuracies": 0.5,
|
| 584 |
+
"rewards/chosen": -0.10037201642990112,
|
| 585 |
+
"rewards/margins": 0.006189983803778887,
|
| 586 |
+
"rewards/rejected": -0.10656200349330902,
|
| 587 |
+
"step": 30
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 2.48,
|
| 591 |
+
"grad_norm": 1.4941612482070923,
|
| 592 |
+
"learning_rate": 4.64e-06,
|
| 593 |
+
"log_odds_chosen": 0.3264733850955963,
|
| 594 |
+
"log_odds_ratio": -0.5566755533218384,
|
| 595 |
+
"logits/chosen": -2.98291015625,
|
| 596 |
+
"logits/rejected": -2.946389675140381,
|
| 597 |
+
"logps/chosen": -0.6631190776824951,
|
| 598 |
+
"logps/rejected": -0.8497029542922974,
|
| 599 |
+
"loss": 0.9711,
|
| 600 |
+
"nll_loss": 0.915428876876831,
|
| 601 |
+
"rewards/accuracies": 0.625,
|
| 602 |
+
"rewards/chosen": -0.06631191074848175,
|
| 603 |
+
"rewards/margins": 0.018658379092812538,
|
| 604 |
+
"rewards/rejected": -0.08497028797864914,
|
| 605 |
+
"step": 31
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 2.56,
|
| 609 |
+
"grad_norm": 2.090597629547119,
|
| 610 |
+
"learning_rate": 4.48e-06,
|
| 611 |
+
"log_odds_chosen": 0.07261727005243301,
|
| 612 |
+
"log_odds_ratio": -0.684439480304718,
|
| 613 |
+
"logits/chosen": -3.051530361175537,
|
| 614 |
+
"logits/rejected": -3.0167644023895264,
|
| 615 |
+
"logps/chosen": -0.861786425113678,
|
| 616 |
+
"logps/rejected": -0.8696616291999817,
|
| 617 |
+
"loss": 1.0085,
|
| 618 |
+
"nll_loss": 0.9400495290756226,
|
| 619 |
+
"rewards/accuracies": 0.5,
|
| 620 |
+
"rewards/chosen": -0.08617864549160004,
|
| 621 |
+
"rewards/margins": 0.0007875119335949421,
|
| 622 |
+
"rewards/rejected": -0.08696617186069489,
|
| 623 |
+
"step": 32
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 2.64,
|
| 627 |
+
"grad_norm": 1.5811887979507446,
|
| 628 |
+
"learning_rate": 4.32e-06,
|
| 629 |
+
"log_odds_chosen": 0.12410806119441986,
|
| 630 |
+
"log_odds_ratio": -0.6687257289886475,
|
| 631 |
+
"logits/chosen": -2.9889068603515625,
|
| 632 |
+
"logits/rejected": -2.7933883666992188,
|
| 633 |
+
"logps/chosen": -1.034374713897705,
|
| 634 |
+
"logps/rejected": -1.1076936721801758,
|
| 635 |
+
"loss": 1.2432,
|
| 636 |
+
"nll_loss": 1.176369071006775,
|
| 637 |
+
"rewards/accuracies": 0.5,
|
| 638 |
+
"rewards/chosen": -0.10343748331069946,
|
| 639 |
+
"rewards/margins": 0.007331886328756809,
|
| 640 |
+
"rewards/rejected": -0.1107693687081337,
|
| 641 |
+
"step": 33
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 2.7199999999999998,
|
| 645 |
+
"grad_norm": 2.2703254222869873,
|
| 646 |
+
"learning_rate": 4.16e-06,
|
| 647 |
+
"log_odds_chosen": 0.12302179634571075,
|
| 648 |
+
"log_odds_ratio": -0.6416699290275574,
|
| 649 |
+
"logits/chosen": -3.0783886909484863,
|
| 650 |
+
"logits/rejected": -3.0010714530944824,
|
| 651 |
+
"logps/chosen": -0.9351593255996704,
|
| 652 |
+
"logps/rejected": -1.035468339920044,
|
| 653 |
+
"loss": 1.1554,
|
| 654 |
+
"nll_loss": 1.0912635326385498,
|
| 655 |
+
"rewards/accuracies": 0.75,
|
| 656 |
+
"rewards/chosen": -0.09351593255996704,
|
| 657 |
+
"rewards/margins": 0.010030903853476048,
|
| 658 |
+
"rewards/rejected": -0.10354683548212051,
|
| 659 |
+
"step": 34
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 2.8,
|
| 663 |
+
"grad_norm": 2.956712484359741,
|
| 664 |
+
"learning_rate": 4e-06,
|
| 665 |
+
"log_odds_chosen": 0.0465146005153656,
|
| 666 |
+
"log_odds_ratio": -0.6841952800750732,
|
| 667 |
+
"logits/chosen": -3.121398448944092,
|
| 668 |
+
"logits/rejected": -3.0879604816436768,
|
| 669 |
+
"logps/chosen": -0.7368811964988708,
|
| 670 |
+
"logps/rejected": -0.7471375465393066,
|
| 671 |
+
"loss": 1.2526,
|
| 672 |
+
"nll_loss": 1.1842174530029297,
|
| 673 |
+
"rewards/accuracies": 0.625,
|
| 674 |
+
"rewards/chosen": -0.07368811964988708,
|
| 675 |
+
"rewards/margins": 0.0010256306268274784,
|
| 676 |
+
"rewards/rejected": -0.07471375167369843,
|
| 677 |
+
"step": 35
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 2.88,
|
| 681 |
+
"grad_norm": 1.5912374258041382,
|
| 682 |
+
"learning_rate": 3.84e-06,
|
| 683 |
+
"log_odds_chosen": 0.011004272848367691,
|
| 684 |
+
"log_odds_ratio": -0.6958502531051636,
|
| 685 |
+
"logits/chosen": -3.0698819160461426,
|
| 686 |
+
"logits/rejected": -3.0204977989196777,
|
| 687 |
+
"logps/chosen": -0.8866332769393921,
|
| 688 |
+
"logps/rejected": -0.892663836479187,
|
| 689 |
+
"loss": 1.1397,
|
| 690 |
+
"nll_loss": 1.0701532363891602,
|
| 691 |
+
"rewards/accuracies": 0.375,
|
| 692 |
+
"rewards/chosen": -0.08866333961486816,
|
| 693 |
+
"rewards/margins": 0.0006030518561601639,
|
| 694 |
+
"rewards/rejected": -0.08926638960838318,
|
| 695 |
+
"step": 36
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 2.88,
|
| 699 |
+
"eval_log_odds_chosen": 0.06313765048980713,
|
| 700 |
+
"eval_log_odds_ratio": -0.6620765328407288,
|
| 701 |
+
"eval_logits/chosen": -3.3640849590301514,
|
| 702 |
+
"eval_logits/rejected": -3.0697708129882812,
|
| 703 |
+
"eval_logps/chosen": -0.661213755607605,
|
| 704 |
+
"eval_logps/rejected": -0.6922562718391418,
|
| 705 |
+
"eval_loss": 0.9372102618217468,
|
| 706 |
+
"eval_nll_loss": 0.8710026144981384,
|
| 707 |
+
"eval_rewards/accuracies": 1.0,
|
| 708 |
+
"eval_rewards/chosen": -0.06612137705087662,
|
| 709 |
+
"eval_rewards/margins": 0.0031042546033859253,
|
| 710 |
+
"eval_rewards/rejected": -0.06922563165426254,
|
| 711 |
+
"eval_runtime": 4.1385,
|
| 712 |
+
"eval_samples_per_second": 0.242,
|
| 713 |
+
"eval_steps_per_second": 0.242,
|
| 714 |
+
"step": 36
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 2.96,
|
| 718 |
+
"grad_norm": 2.163959503173828,
|
| 719 |
+
"learning_rate": 3.68e-06,
|
| 720 |
+
"log_odds_chosen": 0.1411035656929016,
|
| 721 |
+
"log_odds_ratio": -0.6452722549438477,
|
| 722 |
+
"logits/chosen": -2.993727684020996,
|
| 723 |
+
"logits/rejected": -2.9780263900756836,
|
| 724 |
+
"logps/chosen": -0.691759467124939,
|
| 725 |
+
"logps/rejected": -0.7671821117401123,
|
| 726 |
+
"loss": 1.3539,
|
| 727 |
+
"nll_loss": 1.2893797159194946,
|
| 728 |
+
"rewards/accuracies": 0.625,
|
| 729 |
+
"rewards/chosen": -0.06917594373226166,
|
| 730 |
+
"rewards/margins": 0.007542273961007595,
|
| 731 |
+
"rewards/rejected": -0.07671821117401123,
|
| 732 |
+
"step": 37
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 3.04,
|
| 736 |
+
"grad_norm": 5.024839878082275,
|
| 737 |
+
"learning_rate": 3.5199999999999998e-06,
|
| 738 |
+
"log_odds_chosen": 0.24200591444969177,
|
| 739 |
+
"log_odds_ratio": -0.6079150438308716,
|
| 740 |
+
"logits/chosen": -3.187926769256592,
|
| 741 |
+
"logits/rejected": -3.077181816101074,
|
| 742 |
+
"logps/chosen": -1.1563198566436768,
|
| 743 |
+
"logps/rejected": -1.3247044086456299,
|
| 744 |
+
"loss": 1.3491,
|
| 745 |
+
"nll_loss": 1.2883188724517822,
|
| 746 |
+
"rewards/accuracies": 0.875,
|
| 747 |
+
"rewards/chosen": -0.11563197523355484,
|
| 748 |
+
"rewards/margins": 0.016838468611240387,
|
| 749 |
+
"rewards/rejected": -0.13247044384479523,
|
| 750 |
+
"step": 38
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 3.12,
|
| 754 |
+
"grad_norm": 1.4955006837844849,
|
| 755 |
+
"learning_rate": 3.3599999999999996e-06,
|
| 756 |
+
"log_odds_chosen": 0.12877097725868225,
|
| 757 |
+
"log_odds_ratio": -0.6657677292823792,
|
| 758 |
+
"logits/chosen": -2.92303466796875,
|
| 759 |
+
"logits/rejected": -2.85076642036438,
|
| 760 |
+
"logps/chosen": -0.8132851719856262,
|
| 761 |
+
"logps/rejected": -0.8765811920166016,
|
| 762 |
+
"loss": 1.188,
|
| 763 |
+
"nll_loss": 1.1214345693588257,
|
| 764 |
+
"rewards/accuracies": 0.5,
|
| 765 |
+
"rewards/chosen": -0.08132852613925934,
|
| 766 |
+
"rewards/margins": 0.006329594645649195,
|
| 767 |
+
"rewards/rejected": -0.0876581147313118,
|
| 768 |
+
"step": 39
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 3.2,
|
| 772 |
+
"grad_norm": 2.0386250019073486,
|
| 773 |
+
"learning_rate": 3.2e-06,
|
| 774 |
+
"log_odds_chosen": 0.4739803671836853,
|
| 775 |
+
"log_odds_ratio": -0.528855562210083,
|
| 776 |
+
"logits/chosen": -2.883831739425659,
|
| 777 |
+
"logits/rejected": -2.7613730430603027,
|
| 778 |
+
"logps/chosen": -0.5880856513977051,
|
| 779 |
+
"logps/rejected": -0.8737905025482178,
|
| 780 |
+
"loss": 1.0372,
|
| 781 |
+
"nll_loss": 0.9843271374702454,
|
| 782 |
+
"rewards/accuracies": 0.75,
|
| 783 |
+
"rewards/chosen": -0.05880856513977051,
|
| 784 |
+
"rewards/margins": 0.02857048809528351,
|
| 785 |
+
"rewards/rejected": -0.08737905323505402,
|
| 786 |
+
"step": 40
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 3.2800000000000002,
|
| 790 |
+
"grad_norm": 1.866861343383789,
|
| 791 |
+
"learning_rate": 3.0399999999999997e-06,
|
| 792 |
+
"log_odds_chosen": 0.175949364900589,
|
| 793 |
+
"log_odds_ratio": -0.6121445894241333,
|
| 794 |
+
"logits/chosen": -2.689525842666626,
|
| 795 |
+
"logits/rejected": -2.5920565128326416,
|
| 796 |
+
"logps/chosen": -0.8869262933731079,
|
| 797 |
+
"logps/rejected": -1.0022177696228027,
|
| 798 |
+
"loss": 1.0809,
|
| 799 |
+
"nll_loss": 1.0196477174758911,
|
| 800 |
+
"rewards/accuracies": 0.75,
|
| 801 |
+
"rewards/chosen": -0.08869262784719467,
|
| 802 |
+
"rewards/margins": 0.011529151350259781,
|
| 803 |
+
"rewards/rejected": -0.10022178292274475,
|
| 804 |
+
"step": 41
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.36,
|
| 808 |
+
"grad_norm": 1.7331321239471436,
|
| 809 |
+
"learning_rate": 2.88e-06,
|
| 810 |
+
"log_odds_chosen": 0.21247801184654236,
|
| 811 |
+
"log_odds_ratio": -0.6190555691719055,
|
| 812 |
+
"logits/chosen": -2.9846858978271484,
|
| 813 |
+
"logits/rejected": -3.0182762145996094,
|
| 814 |
+
"logps/chosen": -0.7801945209503174,
|
| 815 |
+
"logps/rejected": -0.904188871383667,
|
| 816 |
+
"loss": 1.1836,
|
| 817 |
+
"nll_loss": 1.1216562986373901,
|
| 818 |
+
"rewards/accuracies": 0.75,
|
| 819 |
+
"rewards/chosen": -0.07801946252584457,
|
| 820 |
+
"rewards/margins": 0.012399429455399513,
|
| 821 |
+
"rewards/rejected": -0.09041889011859894,
|
| 822 |
+
"step": 42
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 3.44,
|
| 826 |
+
"grad_norm": 1.5410895347595215,
|
| 827 |
+
"learning_rate": 2.7200000000000002e-06,
|
| 828 |
+
"log_odds_chosen": 0.16515614092350006,
|
| 829 |
+
"log_odds_ratio": -0.6415831446647644,
|
| 830 |
+
"logits/chosen": -3.103332996368408,
|
| 831 |
+
"logits/rejected": -3.0033347606658936,
|
| 832 |
+
"logps/chosen": -0.6259253025054932,
|
| 833 |
+
"logps/rejected": -0.6716697216033936,
|
| 834 |
+
"loss": 0.9397,
|
| 835 |
+
"nll_loss": 0.8755762577056885,
|
| 836 |
+
"rewards/accuracies": 0.375,
|
| 837 |
+
"rewards/chosen": -0.0625925362110138,
|
| 838 |
+
"rewards/margins": 0.004574435297399759,
|
| 839 |
+
"rewards/rejected": -0.06716696918010712,
|
| 840 |
+
"step": 43
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 3.52,
|
| 844 |
+
"grad_norm": 2.591630697250366,
|
| 845 |
+
"learning_rate": 2.56e-06,
|
| 846 |
+
"log_odds_chosen": 0.25676825642585754,
|
| 847 |
+
"log_odds_ratio": -0.6107382774353027,
|
| 848 |
+
"logits/chosen": -3.0657176971435547,
|
| 849 |
+
"logits/rejected": -2.972097873687744,
|
| 850 |
+
"logps/chosen": -0.8358463048934937,
|
| 851 |
+
"logps/rejected": -0.9837089776992798,
|
| 852 |
+
"loss": 0.9461,
|
| 853 |
+
"nll_loss": 0.8850574493408203,
|
| 854 |
+
"rewards/accuracies": 0.625,
|
| 855 |
+
"rewards/chosen": -0.08358462154865265,
|
| 856 |
+
"rewards/margins": 0.014786271378397942,
|
| 857 |
+
"rewards/rejected": -0.09837089478969574,
|
| 858 |
+
"step": 44
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 3.6,
|
| 862 |
+
"grad_norm": 3.252107858657837,
|
| 863 |
+
"learning_rate": 2.4e-06,
|
| 864 |
+
"log_odds_chosen": 0.2689153254032135,
|
| 865 |
+
"log_odds_ratio": -0.59128737449646,
|
| 866 |
+
"logits/chosen": -3.036353826522827,
|
| 867 |
+
"logits/rejected": -2.9656076431274414,
|
| 868 |
+
"logps/chosen": -0.8418894410133362,
|
| 869 |
+
"logps/rejected": -0.9865555167198181,
|
| 870 |
+
"loss": 1.1173,
|
| 871 |
+
"nll_loss": 1.0581891536712646,
|
| 872 |
+
"rewards/accuracies": 0.625,
|
| 873 |
+
"rewards/chosen": -0.08418895304203033,
|
| 874 |
+
"rewards/margins": 0.014466602355241776,
|
| 875 |
+
"rewards/rejected": -0.09865555167198181,
|
| 876 |
+
"step": 45
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 3.68,
|
| 880 |
+
"grad_norm": 2.0005674362182617,
|
| 881 |
+
"learning_rate": 2.24e-06,
|
| 882 |
+
"log_odds_chosen": 0.08924929052591324,
|
| 883 |
+
"log_odds_ratio": -0.6686105728149414,
|
| 884 |
+
"logits/chosen": -3.0594611167907715,
|
| 885 |
+
"logits/rejected": -3.006617307662964,
|
| 886 |
+
"logps/chosen": -0.9009665250778198,
|
| 887 |
+
"logps/rejected": -0.9710016250610352,
|
| 888 |
+
"loss": 1.2238,
|
| 889 |
+
"nll_loss": 1.1569262742996216,
|
| 890 |
+
"rewards/accuracies": 0.625,
|
| 891 |
+
"rewards/chosen": -0.09009665250778198,
|
| 892 |
+
"rewards/margins": 0.00700351083651185,
|
| 893 |
+
"rewards/rejected": -0.0971001610159874,
|
| 894 |
+
"step": 46
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 3.76,
|
| 898 |
+
"grad_norm": 1.9500302076339722,
|
| 899 |
+
"learning_rate": 2.08e-06,
|
| 900 |
+
"log_odds_chosen": -0.01365312933921814,
|
| 901 |
+
"log_odds_ratio": -0.7387136220932007,
|
| 902 |
+
"logits/chosen": -3.102898120880127,
|
| 903 |
+
"logits/rejected": -3.0505154132843018,
|
| 904 |
+
"logps/chosen": -0.9508839845657349,
|
| 905 |
+
"logps/rejected": -0.9444030523300171,
|
| 906 |
+
"loss": 1.2465,
|
| 907 |
+
"nll_loss": 1.1726138591766357,
|
| 908 |
+
"rewards/accuracies": 0.5,
|
| 909 |
+
"rewards/chosen": -0.0950884073972702,
|
| 910 |
+
"rewards/margins": -0.0006480957381427288,
|
| 911 |
+
"rewards/rejected": -0.09444031119346619,
|
| 912 |
+
"step": 47
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 3.84,
|
| 916 |
+
"grad_norm": 1.4190654754638672,
|
| 917 |
+
"learning_rate": 1.92e-06,
|
| 918 |
+
"log_odds_chosen": 0.19316454231739044,
|
| 919 |
+
"log_odds_ratio": -0.6173151731491089,
|
| 920 |
+
"logits/chosen": -3.0933785438537598,
|
| 921 |
+
"logits/rejected": -3.019509792327881,
|
| 922 |
+
"logps/chosen": -0.7028050422668457,
|
| 923 |
+
"logps/rejected": -0.811437726020813,
|
| 924 |
+
"loss": 1.1042,
|
| 925 |
+
"nll_loss": 1.0424754619598389,
|
| 926 |
+
"rewards/accuracies": 0.5,
|
| 927 |
+
"rewards/chosen": -0.07028050720691681,
|
| 928 |
+
"rewards/margins": 0.01086327526718378,
|
| 929 |
+
"rewards/rejected": -0.08114377409219742,
|
| 930 |
+
"step": 48
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 3.84,
|
| 934 |
+
"eval_log_odds_chosen": 0.036497294902801514,
|
| 935 |
+
"eval_log_odds_ratio": -0.6750649809837341,
|
| 936 |
+
"eval_logits/chosen": -3.3623147010803223,
|
| 937 |
+
"eval_logits/rejected": -3.080049991607666,
|
| 938 |
+
"eval_logps/chosen": -0.6492528319358826,
|
| 939 |
+
"eval_logps/rejected": -0.6668488383293152,
|
| 940 |
+
"eval_loss": 0.9257785677909851,
|
| 941 |
+
"eval_nll_loss": 0.8582720756530762,
|
| 942 |
+
"eval_rewards/accuracies": 1.0,
|
| 943 |
+
"eval_rewards/chosen": -0.06492528319358826,
|
| 944 |
+
"eval_rewards/margins": 0.0017596036195755005,
|
| 945 |
+
"eval_rewards/rejected": -0.06668488681316376,
|
| 946 |
+
"eval_runtime": 4.1545,
|
| 947 |
+
"eval_samples_per_second": 0.241,
|
| 948 |
+
"eval_steps_per_second": 0.241,
|
| 949 |
+
"step": 48
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 3.92,
|
| 953 |
+
"grad_norm": 2.1789004802703857,
|
| 954 |
+
"learning_rate": 1.7599999999999999e-06,
|
| 955 |
+
"log_odds_chosen": -0.04005882143974304,
|
| 956 |
+
"log_odds_ratio": -0.7334250211715698,
|
| 957 |
+
"logits/chosen": -2.9112021923065186,
|
| 958 |
+
"logits/rejected": -2.9076766967773438,
|
| 959 |
+
"logps/chosen": -0.765835165977478,
|
| 960 |
+
"logps/rejected": -0.7128950357437134,
|
| 961 |
+
"loss": 1.2473,
|
| 962 |
+
"nll_loss": 1.1739940643310547,
|
| 963 |
+
"rewards/accuracies": 0.375,
|
| 964 |
+
"rewards/chosen": -0.07658351957798004,
|
| 965 |
+
"rewards/margins": -0.005294016562402248,
|
| 966 |
+
"rewards/rejected": -0.07128950208425522,
|
| 967 |
+
"step": 49
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 4.0,
|
| 971 |
+
"grad_norm": 2.178314447402954,
|
| 972 |
+
"learning_rate": 1.6e-06,
|
| 973 |
+
"log_odds_chosen": -0.07708044350147247,
|
| 974 |
+
"log_odds_ratio": -0.7379593253135681,
|
| 975 |
+
"logits/chosen": -3.1038739681243896,
|
| 976 |
+
"logits/rejected": -3.0927176475524902,
|
| 977 |
+
"logps/chosen": -0.9102978706359863,
|
| 978 |
+
"logps/rejected": -0.8448383808135986,
|
| 979 |
+
"loss": 1.2398,
|
| 980 |
+
"nll_loss": 1.1660352945327759,
|
| 981 |
+
"rewards/accuracies": 0.375,
|
| 982 |
+
"rewards/chosen": -0.09102979302406311,
|
| 983 |
+
"rewards/margins": -0.006545952521264553,
|
| 984 |
+
"rewards/rejected": -0.08448383212089539,
|
| 985 |
+
"step": 50
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 4.08,
|
| 989 |
+
"grad_norm": 1.5347027778625488,
|
| 990 |
+
"learning_rate": 1.44e-06,
|
| 991 |
+
"log_odds_chosen": 0.10280649363994598,
|
| 992 |
+
"log_odds_ratio": -0.6480928063392639,
|
| 993 |
+
"logits/chosen": -2.852008581161499,
|
| 994 |
+
"logits/rejected": -2.8604021072387695,
|
| 995 |
+
"logps/chosen": -0.6744309067726135,
|
| 996 |
+
"logps/rejected": -0.7366931438446045,
|
| 997 |
+
"loss": 1.0266,
|
| 998 |
+
"nll_loss": 0.9618306756019592,
|
| 999 |
+
"rewards/accuracies": 0.5,
|
| 1000 |
+
"rewards/chosen": -0.06744308769702911,
|
| 1001 |
+
"rewards/margins": 0.00622622249647975,
|
| 1002 |
+
"rewards/rejected": -0.07366931438446045,
|
| 1003 |
+
"step": 51
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 4.16,
|
| 1007 |
+
"grad_norm": 1.4483627080917358,
|
| 1008 |
+
"learning_rate": 1.28e-06,
|
| 1009 |
+
"log_odds_chosen": 0.07863382995128632,
|
| 1010 |
+
"log_odds_ratio": -0.6696581840515137,
|
| 1011 |
+
"logits/chosen": -3.0924391746520996,
|
| 1012 |
+
"logits/rejected": -3.02586030960083,
|
| 1013 |
+
"logps/chosen": -0.7803107500076294,
|
| 1014 |
+
"logps/rejected": -0.8022903203964233,
|
| 1015 |
+
"loss": 1.0676,
|
| 1016 |
+
"nll_loss": 1.0006568431854248,
|
| 1017 |
+
"rewards/accuracies": 0.375,
|
| 1018 |
+
"rewards/chosen": -0.07803107798099518,
|
| 1019 |
+
"rewards/margins": 0.002197958528995514,
|
| 1020 |
+
"rewards/rejected": -0.08022903650999069,
|
| 1021 |
+
"step": 52
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 4.24,
|
| 1025 |
+
"grad_norm": 2.0634541511535645,
|
| 1026 |
+
"learning_rate": 1.12e-06,
|
| 1027 |
+
"log_odds_chosen": 0.4366196393966675,
|
| 1028 |
+
"log_odds_ratio": -0.5555592775344849,
|
| 1029 |
+
"logits/chosen": -2.8958981037139893,
|
| 1030 |
+
"logits/rejected": -2.7654147148132324,
|
| 1031 |
+
"logps/chosen": -0.9317148923873901,
|
| 1032 |
+
"logps/rejected": -1.197984218597412,
|
| 1033 |
+
"loss": 1.0649,
|
| 1034 |
+
"nll_loss": 1.0093390941619873,
|
| 1035 |
+
"rewards/accuracies": 0.5,
|
| 1036 |
+
"rewards/chosen": -0.09317149221897125,
|
| 1037 |
+
"rewards/margins": 0.026626940816640854,
|
| 1038 |
+
"rewards/rejected": -0.1197984367609024,
|
| 1039 |
+
"step": 53
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 4.32,
|
| 1043 |
+
"grad_norm": 1.9966628551483154,
|
| 1044 |
+
"learning_rate": 9.6e-07,
|
| 1045 |
+
"log_odds_chosen": 0.22570586204528809,
|
| 1046 |
+
"log_odds_ratio": -0.598256528377533,
|
| 1047 |
+
"logits/chosen": -3.159971237182617,
|
| 1048 |
+
"logits/rejected": -3.048675298690796,
|
| 1049 |
+
"logps/chosen": -0.7773997187614441,
|
| 1050 |
+
"logps/rejected": -0.9069120287895203,
|
| 1051 |
+
"loss": 1.0975,
|
| 1052 |
+
"nll_loss": 1.0376590490341187,
|
| 1053 |
+
"rewards/accuracies": 0.625,
|
| 1054 |
+
"rewards/chosen": -0.07773996889591217,
|
| 1055 |
+
"rewards/margins": 0.012951226904988289,
|
| 1056 |
+
"rewards/rejected": -0.09069120138883591,
|
| 1057 |
+
"step": 54
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 4.4,
|
| 1061 |
+
"grad_norm": 1.502306342124939,
|
| 1062 |
+
"learning_rate": 8e-07,
|
| 1063 |
+
"log_odds_chosen": 0.04895332455635071,
|
| 1064 |
+
"log_odds_ratio": -0.7371360659599304,
|
| 1065 |
+
"logits/chosen": -3.0577011108398438,
|
| 1066 |
+
"logits/rejected": -2.9929468631744385,
|
| 1067 |
+
"logps/chosen": -0.8072652816772461,
|
| 1068 |
+
"logps/rejected": -0.8173755407333374,
|
| 1069 |
+
"loss": 1.0562,
|
| 1070 |
+
"nll_loss": 0.9824643135070801,
|
| 1071 |
+
"rewards/accuracies": 0.5,
|
| 1072 |
+
"rewards/chosen": -0.08072653412818909,
|
| 1073 |
+
"rewards/margins": 0.0010110288858413696,
|
| 1074 |
+
"rewards/rejected": -0.08173755556344986,
|
| 1075 |
+
"step": 55
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 4.48,
|
| 1079 |
+
"grad_norm": 1.947930097579956,
|
| 1080 |
+
"learning_rate": 6.4e-07,
|
| 1081 |
+
"log_odds_chosen": 0.16152703762054443,
|
| 1082 |
+
"log_odds_ratio": -0.633980929851532,
|
| 1083 |
+
"logits/chosen": -3.028554916381836,
|
| 1084 |
+
"logits/rejected": -2.980391025543213,
|
| 1085 |
+
"logps/chosen": -0.8234447836875916,
|
| 1086 |
+
"logps/rejected": -0.9135499000549316,
|
| 1087 |
+
"loss": 1.1579,
|
| 1088 |
+
"nll_loss": 1.0945292711257935,
|
| 1089 |
+
"rewards/accuracies": 0.75,
|
| 1090 |
+
"rewards/chosen": -0.08234447985887527,
|
| 1091 |
+
"rewards/margins": 0.009010514244437218,
|
| 1092 |
+
"rewards/rejected": -0.09135499596595764,
|
| 1093 |
+
"step": 56
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 4.5600000000000005,
|
| 1097 |
+
"grad_norm": 1.6086125373840332,
|
| 1098 |
+
"learning_rate": 4.8e-07,
|
| 1099 |
+
"log_odds_chosen": -0.02157880738377571,
|
| 1100 |
+
"log_odds_ratio": -0.7184982895851135,
|
| 1101 |
+
"logits/chosen": -2.8279547691345215,
|
| 1102 |
+
"logits/rejected": -2.8562469482421875,
|
| 1103 |
+
"logps/chosen": -0.8981537222862244,
|
| 1104 |
+
"logps/rejected": -0.8426607847213745,
|
| 1105 |
+
"loss": 1.1922,
|
| 1106 |
+
"nll_loss": 1.1203769445419312,
|
| 1107 |
+
"rewards/accuracies": 0.625,
|
| 1108 |
+
"rewards/chosen": -0.08981536328792572,
|
| 1109 |
+
"rewards/margins": -0.005549294408410788,
|
| 1110 |
+
"rewards/rejected": -0.08426607400178909,
|
| 1111 |
+
"step": 57
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 4.64,
|
| 1115 |
+
"grad_norm": 2.1013498306274414,
|
| 1116 |
+
"learning_rate": 3.2e-07,
|
| 1117 |
+
"log_odds_chosen": 0.09676241874694824,
|
| 1118 |
+
"log_odds_ratio": -0.6955091953277588,
|
| 1119 |
+
"logits/chosen": -2.927985668182373,
|
| 1120 |
+
"logits/rejected": -2.8649566173553467,
|
| 1121 |
+
"logps/chosen": -0.7449172735214233,
|
| 1122 |
+
"logps/rejected": -0.7649428844451904,
|
| 1123 |
+
"loss": 1.1142,
|
| 1124 |
+
"nll_loss": 1.0446666479110718,
|
| 1125 |
+
"rewards/accuracies": 0.5,
|
| 1126 |
+
"rewards/chosen": -0.07449173182249069,
|
| 1127 |
+
"rewards/margins": 0.0020025642588734627,
|
| 1128 |
+
"rewards/rejected": -0.07649429142475128,
|
| 1129 |
+
"step": 58
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 4.72,
|
| 1133 |
+
"grad_norm": 1.8846663236618042,
|
| 1134 |
+
"learning_rate": 1.6e-07,
|
| 1135 |
+
"log_odds_chosen": 0.09952077269554138,
|
| 1136 |
+
"log_odds_ratio": -0.65927654504776,
|
| 1137 |
+
"logits/chosen": -3.1487090587615967,
|
| 1138 |
+
"logits/rejected": -3.0312399864196777,
|
| 1139 |
+
"logps/chosen": -0.8334336280822754,
|
| 1140 |
+
"logps/rejected": -0.9005411267280579,
|
| 1141 |
+
"loss": 1.0835,
|
| 1142 |
+
"nll_loss": 1.0175987482070923,
|
| 1143 |
+
"rewards/accuracies": 0.75,
|
| 1144 |
+
"rewards/chosen": -0.08334335684776306,
|
| 1145 |
+
"rewards/margins": 0.006710751913487911,
|
| 1146 |
+
"rewards/rejected": -0.09005410969257355,
|
| 1147 |
+
"step": 59
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 4.8,
|
| 1151 |
+
"grad_norm": 1.5581363439559937,
|
| 1152 |
+
"learning_rate": 0.0,
|
| 1153 |
+
"log_odds_chosen": 0.30142536759376526,
|
| 1154 |
+
"log_odds_ratio": -0.5821777582168579,
|
| 1155 |
+
"logits/chosen": -3.118028163909912,
|
| 1156 |
+
"logits/rejected": -3.039905309677124,
|
| 1157 |
+
"logps/chosen": -0.6326413154602051,
|
| 1158 |
+
"logps/rejected": -0.7793558239936829,
|
| 1159 |
+
"loss": 1.0224,
|
| 1160 |
+
"nll_loss": 0.964179277420044,
|
| 1161 |
+
"rewards/accuracies": 0.75,
|
| 1162 |
+
"rewards/chosen": -0.06326413154602051,
|
| 1163 |
+
"rewards/margins": 0.01467144675552845,
|
| 1164 |
+
"rewards/rejected": -0.0779355838894844,
|
| 1165 |
+
"step": 60
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 4.8,
|
| 1169 |
+
"eval_log_odds_chosen": 0.03376197814941406,
|
| 1170 |
+
"eval_log_odds_ratio": -0.6764087080955505,
|
| 1171 |
+
"eval_logits/chosen": -3.361872434616089,
|
| 1172 |
+
"eval_logits/rejected": -3.0813393592834473,
|
| 1173 |
+
"eval_logps/chosen": -0.6450251936912537,
|
| 1174 |
+
"eval_logps/rejected": -0.6612161993980408,
|
| 1175 |
+
"eval_loss": 0.9213418960571289,
|
| 1176 |
+
"eval_nll_loss": 0.8537010550498962,
|
| 1177 |
+
"eval_rewards/accuracies": 1.0,
|
| 1178 |
+
"eval_rewards/chosen": -0.0645025223493576,
|
| 1179 |
+
"eval_rewards/margins": 0.001619100570678711,
|
| 1180 |
+
"eval_rewards/rejected": -0.06612162292003632,
|
| 1181 |
+
"eval_runtime": 3.9606,
|
| 1182 |
+
"eval_samples_per_second": 0.252,
|
| 1183 |
+
"eval_steps_per_second": 0.252,
|
| 1184 |
+
"step": 60
|
| 1185 |
+
}
|
| 1186 |
+
],
|
| 1187 |
+
"logging_steps": 1,
|
| 1188 |
+
"max_steps": 60,
|
| 1189 |
+
"num_input_tokens_seen": 0,
|
| 1190 |
+
"num_train_epochs": 5,
|
| 1191 |
+
"save_steps": 500,
|
| 1192 |
+
"stateful_callbacks": {
|
| 1193 |
+
"TrainerControl": {
|
| 1194 |
+
"args": {
|
| 1195 |
+
"should_epoch_stop": false,
|
| 1196 |
+
"should_evaluate": false,
|
| 1197 |
+
"should_log": false,
|
| 1198 |
+
"should_save": true,
|
| 1199 |
+
"should_training_stop": true
|
| 1200 |
+
},
|
| 1201 |
+
"attributes": {}
|
| 1202 |
+
}
|
| 1203 |
+
},
|
| 1204 |
+
"total_flos": 0.0,
|
| 1205 |
+
"train_batch_size": 2,
|
| 1206 |
+
"trial_name": null,
|
| 1207 |
+
"trial_params": null
|
| 1208 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33207cb9fa369eebdfc2da58933dc348fcf0f74eedcd1e0bddfa03326c9366fc
|
| 3 |
+
size 5496
|